SiFuh changed the topic of #crux-social to: Offtopic Talks | Project https://crux.nu/ | Logs: https://libera.irclog.whitequark.org/crux-social/
SiFuh has quit [Remote host closed the connection]
SiFuh has joined #crux-social
serpente has quit [Remote host closed the connection]
LDelaforet has joined #crux-social
<SiFuh> farkuhar: https://www.youtube.com/watch?v=h5wMPDuio78 What the hell is he shouting for? Hah!
SiFuh_ has joined #crux-social
SiFuh has quit [Ping timeout: 272 seconds]
zorz has quit [Ping timeout: 244 seconds]
LDelaforet has quit [Ping timeout: 276 seconds]
LDelaforet has joined #crux-social
LDelaforet has quit [Ping timeout: 245 seconds]
<SiFuh_> farkuhar: Wondering why this Guinness I bought tastes like shit. The realised I picked up the wrong can of Guinness.
<SiFuh_> dlcusa: There are many people standing by looking at the helicopter crash into the Hudson. But I didn't see a single person take off their shoes and jump in to pull the people out of the wreckaged.
<SiFuh_> dlcusa: Also two plans just clipped wings. One taxiing and the other parked.
<SiFuh_> planes*
<farkuhar> SiFuh_: What constitutes a "wrong can of Guinness"?
<SiFuh_> farkuhar: The one on the left is the wrong on. As for constitutes I think I said "tastes like shit" https://i.snipboard.io/cDEsPS.jpg
<SiFuh_> on/one
<SiFuh_> The shit one is 5.5% cool. But it is regular Malaysian Guinness. The one on the right is 4.1%, sad... but tastes way better because it has nitrogen inside.
<farkuhar> I'm not surprised at the onlookers' behaviour at the Hudson helicopter crash. Not many people are eager to play the role of heroic rescuer these days.
<SiFuh_> farkuhar: I would have jumped in. Even if I couldn't swim deep enough in that liquid shit. I still might be able to grab someone if they did manage to exit.
<SiFuh_> And if they were wearing boots. Bloody hell, that is not easy to swim in.
<SiFuh_> They probably would have been trained to swim in boots though.
<SiFuh_> ukky: I went to the shop that sells the fences. The price was originally 934 MYR. But no bolts for the base and no delivery fee. Got it for 800.60 MYR with the bolts for the base and delivery.
<SiFuh_> ukky: It is weird talking to them. I want Mitre bolts. Basically Concrete Anchor Bolts. They call them 'Wall Plugs' here.
LDelaforet has joined #crux-social
LDelaforet has quit [Ping timeout: 276 seconds]
<farkuhar> Hmm, so apparently at least some of my rust problems stemmed from the clang#19.1.7 that I built on March 30. After reinstalling the older clang package from March 13, I got a successful build of rust 1.86 and librsvg.
<SiFuh_> So you only half clanged the gong and didn't bother shouting the partial success?
LDelaforet has joined #crux-social
<farkuhar> partial success, heh. SiFuh_ correctly anticipated that I would still hit problems when updating other rust-dependent ports. Here's firefox, for example: https://dpaste.com/2BMLSZ5FX
<farkuhar> Errors like those were the motivation for trying a new configuration of clang on March 30. It's unfortunate that llvm, clang, compiler-rt have a default configuration that varies from one target platform to the next. Hence the llvm Pkgfile on CRUX-glibc can leave certain settings undeclared, but the Pkgfile in a musl overlay has to toggle those settings explicitly, to get a similarly-provisioned toolchain.
<farkuhar> Compare how emmett1 configured clang 18.1.8 -- https://codeberg.org/emmett1/crux-musl/src/branch/main/clang/Pkgfile -- with how void configured clang 19.1.4 -- https://github.com/void-linux/void-packages/blob/master/srcpkgs/llvm19/template . Void seems to think it's safer to declare more options explicitly, rather than trusting the upstream defaults for the target platform.
<farkuhar> Heh, maybe llvm 18.1.8 _did_ have better defaults when target=x86_64-unknown-linux-musl, and in more recent versions of llvm they switched the defaults, creating a greater divergence between glibc and musl builds. In that case the Void approach is better, because you get a consistent configuration regardless of what the upstream devs think is appropriate for x86_64-unknown-linux-musl.
<farkuhar> It was the March 30 build where I tried the option -DLIBCLANG_BUILD_STATIC=ON, which supposedly resolved an error almost identical to the one I hit with firefox. But that clang option on its own might not be enough; the relevant vendored crate (gecko-profiler) might need to be patched similar to what they did here: https://github.com/rust-rocksdb/rust-rocksdb/pull/939/files
<farkuhar> One of my attempts to fix the failing librsvg build took the ham-fisted approach of deleting the entire line "subdir rsvg_convert" from meson.build, on the theory that the proc-macro crate was unsalvageable. Maybe I could do something similar to prevent the firefox build from descending into the subdir for gecko-profiler (hopefully the absence of this feature won't cause runtime breakage).
LDelaforet has quit [Ping timeout: 248 seconds]